ASUNTOS DE CARACTER EXTRAORDINARIO

In this chapter we have explored the possibility of studying cosmology through radio- telescopes that operate in the redshift range 2.5 < z < 5. The reason behind this is that, while this is a redshift-range not considered in current and upcoming setups, the volume it encloses is much larger then the one probed by current and future spectroscopic surveys. The question we try to answer is: how much cosmological information is contained in this redshift window?

We focus our analysis on four key cosmological quantities: 1) the growth rate, f σ₈, the BAO distance scale parameters, DAand H, the sum of the neutrino masses, Σmν,

and the number of relativistic degrees of freedom, N_eff. We consider four extensions of current or upcoming radio-telescopes like HIRAX, CHIME and FAST, and two different observational strategies: interferometry (for Ext-HIRAX and Ext-CHIME) and single-dish (highzFast).

We carry out our analysis using the Fisher matrix formalism. We model the ampli- tude and shape of the 21cm signal using the model proposed in [76]. We also account for cosmological and instrumental effects such as the presence of the wedge, the win- dow function, the instrument thermal noise, the angular resolution, the presence of shot-noise...etc.

We point out that measurements that are sensitive to the overall amplitude of the 21cm power spectrum, like f σ₈, will be completely degenerate with astrophysical parameters like Ω_HI and b_HI. In order to break that degeneracy it is necessary that we use independent datasets that constrain those quantities. We show how the value of these parameters can be determined through either the Lyα-forest alone or via cross-correlations between 21cm and the Lyα-forest or DLAs.

Under the assumption of the primary beam foreground wedge contamination (mid- wedge case in the text), 5% priors on b_HI and Ω_HI and k_max= 0.2 hMpc−1, that we term the fiducial setup, we find that Ext-HIRAX can constrain the value of f σ₈within bins of ∆z = 0.1 at ' 4% in the redshift range 2.5 < z < 5. A modest improvement is achieved by changing k_max from 0.2 hMpc−1 to k_nl. If data from the whole wedge need to be discarded, these constraints degrade between a factor 2 (at z = 2.5) and 7 (at z = 5).

Under the fiducial setup, we find that Ext-HIRAX will place ' 1% constraints on

DAand H. As with the growth rate, our results point out that going to smaller scales

has only a very modest impact on the results. Being able to use a fraction of the modes in the wedge has a huge impact on our results, as removing the information in the whole wedge degrades the constraints between a factor of 10 (at z = 2.5) and 20 (at z = 5).

We have also studied the impact that the theory model has on the results. By using a theory template that accounts for 1-loop corrections and incorporates 2 free parameters that we marginalise over, we find that the constraints on the cosmological parameters worsen between 10% and 500% when 2% priors on Ω_HI and b_HI are used. In the case of the neutrino masses, the constraints worsen between 10% and 30%.

We find that data from Ext-HIRAX in the fiducial setup can constrain the neutrino masses with an error of 0.10 eV. In combination with data from CMB S4 and galaxy clustering from Euclid the errors shrink to ' 20 meV. Our results are not very sensitive to the wedge coverage, the minimum scale employed and the priors on bHI and ΩHI.

Finally, we find that data from Ext-HIRAX, in the fiducial setup, plus CMB S4 plus Euclid can constrain N_eff with a very competitive error of 0.02. As with the neutrino masses, our constraints do not depend much on the ΩHIand bHI priors, kmax and the wedge coverage.

4.4. Summary and conclusions 79

Results for the Ext-CHIME and highzFAST instruments are similar to those of Ext-HIRAX, with the exception of neutrino masses, where highzFAST performs worse than Ext-CHIME or Ext-HIRAX.

We conclude that there is a large amount of cosmological information embedded in the, poorly constrained, redshift range 2.5 < z < 5. Suitable extensions of existing and upcoming radio-telescopes targeting at this redshift window can provide very tight constraint on key cosmological parameters.

81

Chapter 5

The HI content of dark matter

halos at z ≈ 0 from ALFALFA

Introduction

In this chapter we will use a self-consistent framework to constrain the M_HI-M_h rela- tion using the mass-weighed clustering of HI galaxies detected by the Arecibo Legacy Fast ALFA survey (ALFALFA), as well as their abundance in halos extracted from galaxy groups found in the SDSS galaxy survey. We will also explore the possibility of constraining the shape of the HI profile and the impact of modeling assumptions on our results.

This chapter is organized as follows. In section 5.1 we will summarize the theo- retical framework already described in section 2.1.1that we use to characterize the abundance and clustering of HI. We outline the data employed in this work in section 5.2. The methods used to analyze the data and compare with the theory predictions are illustrated in section5.3. The main results of this work are shown in section 5.4. We discuss the results and summarize the conclusions of this work in section5.5.

5.1

HI halo model

The purpose of this chapter is to constrain the HI-mass-to-halo-mass relation M_HI(M_h) from direct measurements in selected galaxy groups, as well as from the clustering of HI sources. In order to do that we will use the HI halo model described in section 2.1.1. As we saw there the clustering of HI is dominated by its distribution within the halo (i.e. the so-called 1-halo term). Although our constraints will be based solely on the shape of the correlation function on larger scales, we use two different models for the HI density profile, in order to quantify the effect of this assumption on the final results:

• Altered NFW profile: this is the model introduced and used in [113, 114,

103]. and assumes the radial profile of the form:

ρHI(r|Mh) ∝ (r + 3/4rs)−1(r + rs)−2 (5.1)

where rs is the scale radius of the HI cloud, and is related to the halo virial radius R_v(M_h) by the concentration parameter – c_HI(M_h, z) ≡ Rv(Mh)/rs. We follow [219,220] and use a mass-dependent concentration parameter given by:

cHI(Mh, z = 0) = 4 cHI,0  _M h 1011_M  −0.109 . (5.2)

. In section 2.1.2we have found that a similar form of this profile describes the average HI density profile found in numerical simulations very well. For sim- plicity, we do not consider the exponential cutoff at small scales (see equations 2.8 and 2.8) in this chapter.

• Exponential profile: this is the model implemented in [103], and given by

ρHI(r|Mh) ∝ exp (−r/rs), (5.3) In both cases the proportionality factors are automatically fixed by requiring that the HI mass be given by the volume integral of the density profile up to the halo virial radius R_v(M ).

MHI(Mh) = 4π

Z Rv 0

dr r2ρHI(r|Mh). (5.4)

Thus, both profiles are described by one additional free parameter, c_HI,0. The nor- malized HI density profile in Fourier space for the altered NFW profile is given in [103] (see their equation A3), while the exponential profile is simply

uHI(k|Mh) = 1 (1 + k2_r2 s)2 . (5.5) .

As before, for the halo mass function and bias, we use the parametrizations of [104], derived from numerical simulations, but in this chapter we adhere to halo masses defined by a spherical overdensity parameter ∆ = 180

Mh=

4π

3 ρcΩm∆R 3

v. (5.6)

Finally, our basic clustering data vector is the 2D projected correlation, given by the projection of the 3D correlation function along the line of sight. This can be computed directly from the power spectrum as:

Ξ(σ) = Z ∞ −∞ dπ ξ(π, σ) = Z ∞ 0 k dk 2π [PHI,1h(k) + PHI,2h(k)] J0(kσ), (5.7) where J0(x) is the order-0 cylindrical Bessel function. To accelerate the computation of Ξ(σ) we made use of FFTLog [221].

Our theoretical model therefore depends on four free parameters θ = {M0, Mmin, α, cHI,0}. We fix all cosmological parameters to values compatible with the latest ΛCDM con- straints measured by Planck [6] (H₀ = 70 km s−1Mpc−1, Ω_m= 0.3075, n_s = 0.9667,

5.2. Data 83